MON-164 The Non-insulin-based METS-IR Index Combined With Liver Transaminases Improves Prediction Of Fatty Liver And Fibrosis

BACKGROUND: Liver fat content (LFC) evaluation requires histologic examination of liver biopsy specimens, which is invasive and costly. Non-invasive alternatives to assess LFC, which include spectroscopy-MRI (MRI-S) and transient elastography (TE), are limited by cost and complexity. Laboratory-based surrogates to estimate LFC have been developed and include the AST/ALT ratio, Fatty Liver Index (FLI) and NAFLD score. Previous studies have shown that the non-insulin-based METS-IR index is associated with LFC. OBJECTIVE: To evaluate diagnostic performance of METS-IR combined with liver transaminase assessment to estimate LFC. METHODS: We assessed subjects with and without T2D who had a laboratory and MRI-S evaluation. Using linear regression, we developed an estimate of LFC using METS-IR and liver function tests; prediction was improved including sex and T2D using non-linear regression and bootstrap cross-validation (n=2,000). We contrasted the FLI and NAFLD scores with METS-IR to estimate LFC using partial correlation and to detect NAFLD (LFC>5.5% by MRI-S) using areas AUROC. This assessment was validated in a cohort of 197 subjects who underwent TE and a cohort of 57 subjects in whom liver biopsy specimens were collected during bariatric surgery to assess predictive performance for both liver steatosis and fibrosis. RESULTS: In the discovery cohort, we included 56 healthy subjects and 58 with T2D. We modeled the Metabolic Score for Liver Fat Content (METS-LFC) and LFC (r(2)=0.548) as: METS-LFC=(METS-IR*ALT)/(2*AST), Estimated LFC (%)=-3.524+[0.340*METS-LFC]+1.529(T2D)-1.620(Sex). NAFLD was present in 33 subjects (28.9%) of the discovery cohort. The correlation between estimated and observed LFC was higher for METS-LFC (ρ=0.735 95%CI 0.608-0.828) compared to other estimates as was the AUROC (AUC=0.869 95%CI 0.797-0.941). The validation cohort was composed by lean and obese individuals who had METS-LFC and TE; we observed increasing estimated LFC using METS-LFC with TE-derived liver steatosis categories (p<0.001). METS-IR had the better predictive performance for TE-diagnosed steatosis (AUC 0.860 95%CI 0.807-0.912), whilst FLI, METS-IR and METS-LFC had similarly predictive performance for liver fibrosis. Finally, we validated this index in liver biopsy specimens from subjects with average BMI of 45.54±7.47 kg/m2, 33 patients had steatosis and NASH histologic evidence and 10 had biopsy-proven NASH. Compared to biopsy specimens METS-LFC (AUC=0.759, p=0.004) was superior to FLI (AUC=0.615, p=0.201) and the NAFLD score (AUC=0.746, p=0.006) to detect liver steatosis. CONCLUSION: Combined estimation of METS-IR and liver transaminases improves prediction of liver fat content by MRI, TE and biopsy-proven liver steatosis.